Re: [PATCH] s390/kvm: fix diag318 propagation and reset functionality

[Cornelia Huck]

On Tue, 10 Nov 2020 13:03:47 -0500
Collin Walling <walling@linux.ibm.com> wrote:

> On 11/10/20 5:51 AM, Cornelia Huck wrote:
> > On Wed,  4 Nov 2020 13:12:44 -0500
> > Collin Walling <walling@linux.ibm.com> wrote:

> >> diff --git a/target/s390x/kvm.c b/target/s390x/kvm.c
> >> index baa070fdf7..4b2aad009c 100644
> >> --- a/target/s390x/kvm.c
> >> +++ b/target/s390x/kvm.c
> >> @@ -1615,6 +1615,7 @@ static void handle_diag_318(S390CPU *cpu, struct 
> >> kvm_run *run)
> >>  {
> >>      uint64_t reg = (run->s390_sieic.ipa & 0x00f0) >> 4;
> >>      uint64_t diag318_info = run->s.regs.gprs[reg];
> >> +    CPUState *t;
> >>  
> >>      /*
> >>       * DIAG 318 can only be enabled with KVM support. As such, let's
> >> @@ -1622,13 +1623,14 @@ static void handle_diag_318(S390CPU *cpu, struct 
> >> kvm_run *run)
> >>       */
> >>      if (!s390_has_feat(S390_FEAT_DIAG_318)) {
> >>          kvm_s390_program_interrupt(cpu, PGM_SPECIFICATION);
> >> +        return;  
> > 
> > Seems like this return already fixes a bug on its own?  
> 
> Yes. I can split this into a separate patch.

Not really needed, just a remark :)

Do you think you can get out a v2 soon enough so that it can be
included in the next -rc?

> 
> >   
> >>      }
> >>  
> >> -    cpu->env.diag318_info = diag318_info;
> >> -
> >>      if (can_sync_regs(CPU(cpu), KVM_SYNC_DIAG318)) {
> >> -        run->s.regs.diag318 = diag318_info;
> >> -        run->kvm_dirty_regs |= KVM_SYNC_DIAG318;
> >> +        CPU_FOREACH(t) {
> >> +            run_on_cpu(t, s390_do_cpu_set_diag318,
> >> +                       RUN_ON_CPU_HOST_ULONG(diag318_info));
> >> +        }
> >>      }
> >>  }
> >>    
> > 
> >   
> 
>